[00:56] *** pyvpx has joined #arpnetworks [03:16] *** sga0 has quit IRC (Read error: Connection reset by peer) [04:37] *** LT has joined #arpnetworks [09:34] *** novae has quit IRC (Ping timeout: 265 seconds) [09:36] *** LT has quit IRC (Quit: Leaving) [09:37] *** novae has joined #arpnetworks [09:43] *** novae has quit IRC (Ping timeout: 265 seconds) [09:48] *** novae has joined #arpnetworks [09:54] *** sga0 has joined #arpnetworks [11:47] *** bashas has joined #arpnetworks [11:47] hello everyone [11:47] i'm asking for a good windows vps provider [11:47] any suggestions? [11:48] bashas: budget? [11:48] no limits [11:48] but i need a good reviewed one [11:49] with a good spec [11:49] check VPS offers at: http://www.webhostingtalk.com/forumdisplay.php?s=1c5bb91bee9812bf871e252e57078dda&f=104 [11:49] 'good' can mean different things to different people :-) [11:49] yes :) i know [11:49] but anyway, tnx dude for ur help [11:50] checking the forum [11:50] you are better off asking stuff like: so much RAM, managed vs Un managed etc ? Disk Space, Location [11:51] what's managed an unmanaged? a new term to me? btw i'm a web developer [11:52] and i need to host a project, it has few hits daily, but do some processing every few mins using C# threads [11:52] if you have to ask, you probably need managed [11:52] unmanaged is like no software support from the hohst [11:52] you need to take care of all software install and config [11:53] the project aim is to download some html pages and parse them, and extract data from it and put into a sql db [11:53] host will only support underlying network and virtualization, power etc [11:53] unmanaged is a shared windows u mean, right? [11:54] but managed is a full windows machine with total control, am i right [11:54] the box will be shared, you will have total control on your own instance [11:54] admin / root access [11:54] https://www.liquidweb.com/storm/vps.html [11:55] those guys are pretty good [11:55] i know they are shared, one server hosts few Virtual Machines [11:55] right [11:55] anyway got to go [11:55] ok np bro. [11:56] thanks very much [11:56] i'll try to google and understand the deep details of manage and unmanaged [11:59] Managed: The host/provider "manages" the system, installing updates, changing settings, all the server admin stuff for you. [12:00] Unmanaged: You're responsible for maintaining the server, keeping software up to date, etc. [12:00] ie. your home computer is unmanaged. [12:00] Whereas in a corporate environment with corporate IT, your work computer is "managed" [12:00] Hope that clears things up [12:00] That's what she said!! [12:01] brycec: very useful [12:01] thanks dude :) [12:12] *** bashas has quit IRC (Quit: http://www.kiwiirc.com/ - A hand crafted IRC client) [12:17] is someone trying to get rid of windows users? :) [12:18] i suppose if someone asks if they need manager or unmanaged they need managed. [12:18] oh mnathani said that :) [12:30] lol [12:30] I dare say that a whole VPS for a small site is overkill too [12:31] but I'm not too experienced in the windows hosting arena... maybe there are providers that won't allow you to run your own program periodically? [12:31] (Regardless - sounds like something that a LAMP stack could easily do) [12:32] Man - when a SAN runs out of disk space, servers really don't handle that well [12:32] (for loose definitions of SAN and server) [12:34] (KVM machines with NFS-backed storage, to be exact) [12:36] if it's just for a site, azure sites works well enough [12:36] there's other stuff that can run asp in a similar containerized env too [12:37] Azure has a free tier too, as I recall [12:37] yeah, that's how i played with it [12:38] supports code repo autodeployment too (ala heroku) [13:10] i think background stuff could be annoying on shared windows hosting [13:10] i'm paranoid about shared web hosting myself [13:17] background stuff? [13:19] i finished building my "less complicated" fermenter controller today [13:22] he said about c# threads in background or something? [13:23] oh, probably similar to background php processes or whatever [13:25] I figure it's roughly equivalent to a cron job [13:25] yeah [13:25] just executing at some interval [13:26] does shared hosting normally let you do that? [13:27] * hazardous waves [13:27] shared windows hosting is infuriating to deal with [13:27] no idea, haven't ever used windows shared stuff [13:27] i really really dislike shared anything with asp.net [13:27] because of trust level/privs [13:27] and most shared hosts don't like to give full trust [13:28] and for a lot of things, you have to contact the host [13:28] tend to kind of just run it in a vm because effort [13:30] probably similar cost-wise as vm stuff keeps getting cheaper [13:30] the problem is that it tends to restrict your choice of hosts [13:31] because windows licencing is opening schrodinger's dvd case [13:32] (Every shared lunix host I've had lets me schedule cron jobs, fwiw) [13:32] that being said.. i do have a shared hosting account on a linux place i like, because they let me run my own apache/nginx/rails/python server like flask or something, and just let me use a port to reverse proxy it from frontend loadbalancer [13:32] and allow compilers/installing binaries/etc [13:32] but that's literally the only 1 single linux shared host i know that does that, everything else is terrible [13:33] i hate cpanel/plesk/etc with a passion [13:33] cpanel still claims to support python and ruby on rails, which is a massive, hilarious lie [13:34] tbh i feel like the adoption of non-php web languages is partially because of control panels (and a bit on the languages themselves) [13:34] php is drag and drop, and visit url [13:34] that has never been, nor will it in the near future be the case for nodejs, python, ruby, [...] [13:35] and without ssh access, in general, it's borderline impossible to run on shared hosting [13:35] yeah [13:36] PHP apps are stupid easy to deploy [13:36] as dumb as it is, being able to simply upload index.php and be able to visit index.php [13:36] if you try to explain the concept of routes to someone that "just wants a forum" [13:36] many people don't seem to be able to parse the concept of "upload index.php, and you can visit /login" [13:37] so not supporting apache's single-file .htaccess rewrite thing (which is kind of inefficient since it seems to recurse every subdirectory and whatnot) [13:38] as far as most of the people i deal with know, .htaccess is black magic [13:38] That's what she said!! [13:38] the thing that pisses me off the most is that really shitty shared hosting providers are not helping [13:38] like cpanel hosts that lie about having python support [13:38] and then someone tries to "install a python webapp" and it fails miserably [13:40] * hazardous hugs BryceBot [13:40] by python support, they mean the host has python installed [13:40] not that you can use it [13:40] bahaha [13:41] cpanel is "the standard" [13:41] it's pretty disgusting yes. [13:41] Posted 1 year ago [13:41] Why does CPanel continue ignoring requests for supporting Python/Django [13:41] laugh [13:41] i was surprised by how long it took me to get smokeping working with nginx [13:42] nginx with fastcgi is a bit confusing. [13:42] at least for php, nginx + fcgi is 4 lines [13:42] not so much actual cgi-bin applications [13:43] but i like the idea of micro vps with nginx/php/etc all seperate. [13:43] this is perl [13:43] you need to use fcgiwrap [13:43] not so much actual cgi-bin applications <- i've never actually had to run something cgi/perl [13:43] like for web [13:43] i think that was common maybe before i was born but idk [13:43] there's a few ways to plug that support into nginx [13:43] it's just really foreign to me as a concept [13:43] it was easier on ubuntu than arch linux tbh [13:43] mod_perl sucks [13:43] reminds me [13:44] what happened to gentoo [13:44] i used thttpd as a cgi-proxy years ago [13:44] wasn't it all the nerd-rage a decade ago [13:44] behind nginx [13:44] everyone seems to have moved to arch or something from what i can gather [13:44] i think it's still around [13:44] m0unds: i've been using lighttpd for ages, but thought i'd try nginx because nginx is adding spdy support and fastopen support [13:44] fastopen is really easy to enable in nginx. in lighttpd i have a source hack. [13:45] nginx pisses me off with their commercial product offering [13:45] hazardous: why? [13:45] they have a commercial product offering? [13:45] haha [13:45] i need the ability to kick out unresponsive backends for reverse proxying [13:45] m0unds: yeah. [13:45] this is commercial product only [13:45] minimum few thousand a year [13:45] required support contract [13:45] ah [13:45] they're still putting mroe effort into their open source offering than anyone else. [13:45] there is no way i can buy like "just want this module" [13:45] i don't want support [13:45] i just literally want one config option enabled [13:45] i don't need SLA, i don't need tickets [13:45] but that's totally not available to me [13:46] i would happily pay like 50-100/yr just for the ability to use a few extra config options [13:46] but a few thousand i can't afford [13:46] defining backends to reverse proxy is free - doing anything else with them, like checking if they are online, setting randomization, removing or adding backends [13:46] is a $$$$ feature only [13:46] so backends can hang and it can't fix it? [13:46] isn't that spawnfcgi's job? [13:46] i have this problem in that my backends are closed-source binaries that i don't have access to, and occasionally they die [13:47] i would like to run a script if one doesn't return or sends a 5xx [13:47] i would also like to check periodically if they respond [13:47] so now.. i use chinese nginx [13:47] are you in china hazardous ? [13:47] no [13:48] i'm in freedom-land [13:48] canada? [13:48] wow [13:48] close enough [13:48] Obviously kicking the backend can be scripted (kill $pid), so you're just missing the ability to query a specific backend directly? [13:48] brycec: yeh how would he know which one to kill [13:48] brycec: http://tengine.taobao.org/document/http_upstream_check.html [13:49] i wonder what happened to dtrace coming to linux [13:49] http://tengine.taobao.org/document/http_upstream_consistent_hash.html [13:49] i kind of like the idea of tracing normal programs [13:50] these are all bsaically nonexistent in free nginx [13:50] and some kind of autodetection of weird behaviour [13:50] So it kicks it out of the nginx pool [13:50] * mercutio doesn't trust people to not screw up [13:50] * brycec trusts people to screw up [13:50] * hazardous greps mercutio [13:50] heh brycec [13:51] screwing up is normal human behaviour [13:51] error handling is often terrible on unix [13:51] actualyl with software in general [13:51] brycec: even manually removing a backend temporarily in nginx is paid, iirc [13:51] without editing configs [13:51] much less having proactive health checks that let you define "after x failures, drop this out, but readd it if it starts responding later on" [13:51] relayd can do that btw hazardous [13:52] if you have multiple servers [13:52] Yeah - I ran across a "need" (want, really) to do that and found it only came with money [13:52] if you want a laugh, my primary web frontend has 9 VPNs running on top of each other [13:52] but if one out of 4 backends is doing it or something, and it returns 5xx on one of them i dunno how it'd handle that [13:52] One could easily kick/restart the backend with your own script though, but it does nothing for nginx [13:53] if someone wrote a backend handler that also did performance monitoring etc [13:53] they could probably sell it [13:53] capturing the 5xx and identifying where it came from is the thing for me, because i just want to drop it out of my uplinks for a while until my watchdog processes notice and restart it [13:54] tengine's upstream_check thing lets you set things like timeout msec, expecting 200 OK or 302 or something, and checks ssl handshakes [13:54] by default all backends are offline until they pass their first check [13:54] i really wish i could just use stock nginx, i'd be happy to support their efforts and whatnot, but it's just not feasible for anything not business [13:56] what about sticking a wrapper program around your binarys [13:56] define wrapper program [13:57] [they are windows binaries that listen on a port, running on remote hosts] [13:57] runs your cgi script, sees that it behaves [13:57] has an abort if it takes too long, or if it does 5xx. [13:57] i have 1,539 backends [13:57] oh [13:57] lol; [13:57] relayd should handle that then [13:57] That's what she said!! [13:57] the thing is.. it also returns HTTP 200 for failed requests [13:57] which is bsd hmm. [13:57] with error text in the body [13:57] oh [13:58] that's annoying :) [13:58] so i also check if it returns a XML response that contains 'failed_err' [13:58] you could do your own monitoring script and take them out [13:58] there's a few failure states, one of them is named "SHOULD_NEVER_HAPPEN_RACE_CONDITION_TRAP" [13:58] i've encountered that af ew times [13:58] are all the backends the same, you're just load balancing to them? [13:59] load balancing to them across vpn, every 150 or so is a different country, some are domestic-internet-only, some are only accessible from one isp [13:59] tl;dr asia [13:59] i think your situation is complex enough that it warrants coding for it. [14:00] fun thing of the day, i was told that in south korea, reverse dns is illegal unless i have a south korean business licence, and it can take up to 6 months to complete the application process for a single PTR record, and that delegation of NS/PTR/CNAME is not allowed [14:00] wow. [14:00] all PTR requests have to go through KRNIC [14:00] if you've ever wondered why almost no south korean ip has a PTR, well, that might explain it maybe [14:01] china doesn't have PTR either ? [14:01] china occasionally has isps that have applied for PTR for their whole netblock [14:01] none of which have forward dns, most of them completely invalid [14:01] lol [14:01] this is basically standard for all of asia [14:03] my upstream in vietnam assigns datacenter colo, dynamic adsl, dialup, voip, and iptv out of the same subnet [14:03] i can pull a dynamic adsl ip from colo, from standard dhcp pool [14:03] all static IPs have the reverse dns 'dynamic.vdc.vn', and this isn't editable [14:03] they have several /16's that consist entirely of 'dynamic.vdc.vn' exactly, not unique ptr per ip [14:08] theres also the whole "having to bribe customs and shipping officials or else your hardware gets stolen" thing [14:17] wow. [14:18] has anyone tried the new apache httpd? [14:18] err new openbsd httpd [14:19] lol way to goof that :P [14:19] i'm not exactly sure why openbsd is doing their own httpd [14:19] back in thttpd/boa/lighttpd/etc days there were heaps of them [14:19] i jumped from boa to lighttpd myself [14:19] That's what she said!! [14:19] Same reason they do their own everything - to keep full control over it [14:26] openbsd made a httpd? [14:26] yeh based on relayd it seems. [14:26] So more pf.conf-style syntax! [14:26] looks like it's pretty basic [14:26] they also made their own smtpd [14:26] i like relayd. [14:26] httpd is extremely basic at this point [14:26] all i want/need is static file serving from a dir [14:27] OpenSMTPD is pretty great [14:27] sendmailis prettty bad [14:27] opensmtpd is awesome, i wish there debian/ubuntu packages for it [14:27] its in sid but thats about it [14:28] there needs to be linux kernel with openbsd userland :) [14:29] that exists [14:29] it's called windows [14:29] :) [14:32] what's the "in" orchestration tool nowadays [14:32] puppet? ansible? [14:32] it seems to change weekly [14:33] puppet and chef are the big "in" ones from what I can see [14:33] But there's still a lot to be said about using the right tool for the job [14:33] such as ansible in my case [14:36] ansible <3 [14:39] ansible++ [14:42] ansible++ [14:42] ansible has some interesting advantages [14:43] we interviewed them for floss [14:43] along with puppet, chef, cfengine, saltstack... :) [14:44] I needed something to do some basic script-able tasks, without reinventing the wheel myself. Ansible gives me that and a bit more. I was up and running in ~minutes, rather than the overhead of chef or puppet. [14:45] so the nerd portion is ansible [14:45] i don't want massive overhead, custom daemons, etc [15:16] *** erratic has quit IRC (Ping timeout: 258 seconds) [15:30] *** erratic has joined #arpnetworks [15:34] *** mus1cb0x has joined #arpnetworks [15:42] *** mus1cb0x has left "WeeChat 0.4.2" [16:13] hazardous: reading the scrollback, what was that shared linux host you found that let you run your own web server? [17:18] This looks like a really cool position: https://ams-ix.net/about/careers--2/project-stagiair-noc [18:15] you going to apply mnathani ? [18:16] mercutio: if only I met the requirements [18:16] :-) [18:16] i don't think it had many requirements? [18:16] it sounded like it'd be a bit boring though [18:16] Following this project, the internee will follow/collaborate with [18:17] AMS-IX network engineers to work on [18:17] it's bsaically just a testing/scripting job from what i can see [18:19] I suppose they would have some basic on the job training [18:19] it's also probably in amsterdam [18:20] but yeah what they're doing sounds interesting [19:19] i'm trying to add high dpi support to smokeping [19:19] i got the overview looking better.. [19:20] but smokeping isn't saying the size for normal images when you go to the page :( so i can't just half it. [19:22] > isn't saying the size for normal images [19:22] isn't saying it where? [19:22] in the html [19:22] oh [19:22] it's just referencing the made image [19:22] apparently the way to make high dpi work easiest, is you just define the size as half of the actual image size. [19:22] and high dpi displays will show all the detial. [19:22] got it [19:23] you could mess with the image generation I guess [19:23] That's what she said!! [19:23] is it really? [19:23] BryceBot: no [19:23] Oh, okay... I'm sorry. 'you could mess with the image generation I guess' [19:23] i am doubling the rrdtool created image size [19:23] and want to touch smokeping as little as possible. [19:24] That's what she said!! [19:24] BryceBot: no [19:24] Oh, okay... I'm sorry. 'and want to touch smokeping as little as possible.' [19:24] does smokeping just invoke the rrdtool command with some args? [19:24] i think so [19:25] # HIGH DPI [19:25] $xs=$xs/2; [19:25] $ys=$ys/2; [19:25] that in the right place fixes the overview. [19:25] That's what she said!! [19:25] with just doubling the size in the config. [19:25] BryceBot: no [19:25] Oh, okay... I'm sorry. 'that in the right place fixes the overview.' [19:26] because the overview defines the size explicitly in the html? [19:27] it's perl generated html [19:27] and overview does yes. [19:27] but detail doesn't. [19:27] i could try and understand the code more.. [19:33] *** sga0 has quit IRC (Ping timeout: 245 seconds) [19:35] mercutio: I'm editing .pm files in /usr/share/perl5/Smokeping [19:35] but nothing seems to happen [19:35] got it [19:35] what did you do? [19:35] you have to restart fcgiwrap [19:36] i found that $xs{''} is used somewhere else [19:36] and added in width and height with that [19:36] i haven't done much perl [19:36] cool [19:36] what's the difference between $xs and $xs{''} [19:36] i think the smokeping code is messy :) [19:37] http://pastebin.com/tPWXf0an [19:37] curly braces are used to access hash tables? [19:38] * acf_ doesn't know perl either [19:38] http://202.49.71.24:24/smokeping/smokeping.fcgi?target=Wider.google [19:38] that's got it running on it [19:38] now i just need the text size bigger... [19:38] That's what she said!! [19:38] looks super tiny on my monitor :P [19:39] it would be cool if rrdtool rendered svg instead of raster [19:39] the text or graph? [19:39] i think rrdtool can [19:39] i just wanted to start with something [19:39] well, all of it [19:40] you mean the graph is small or the text is small? [19:40] the text mostly [19:40] yeah the text is small for me too [19:40] on my monitor, it's a bit blurry [19:40] the text on the left is a little small, the text in the image is tiny [19:40] probably because of the scaling [19:40] i'm at 1440p [19:40] but yeah the text size needs to be bumped up [19:40] That's what she said!! [19:41] I'm at 1024x768 atm [19:41] do you use 100% zoom? [19:41] oh [19:41] yea [19:41] ouch at 1024x768 acf_ [19:41] so yeah high dpi won't help you at all [19:41] That's what she said!! [19:41] nope :P [19:41] still it should work without high dpi too :) [19:42] it'll still be blurry a little probably [19:42] I just moved down to 1 1080p display, from 2x 1280x1024 + 1x 1080p [19:42] i have 4k+1440p [19:43] Windows or *Nix? [19:43] 1440p on windows and 4k on linux [19:43] chrome sucks on linux at 4k though [19:43] text looks amazing though :) [19:43] consoles are so ... nice... :) [19:43] white banners on right and left? [19:43] huh? [19:43] how many inches on the 4k display? [19:43] 28" [19:44] vs 27" for 1440p [19:44] it hardly looks any bigger tbh [19:44] they came down in cost a lot [19:44] the white regions when you make the browser full screen [19:44] i'm using radeon 7750 for video [19:44] oh i mean scrolling is slow [19:44] and i haven't been able to make url bar big yet [19:45] actually scrolling was slow on 1440p too [19:45] it just got worse [19:45] the video card should be fine... [19:45] i think it's sucky open source drivers. [19:47] I wonder if I just change all the .png s to .svg in smokeping... [19:47] arrg [19:48] so it seems like certain edits to Graphs.pm aren't applying [19:48] I change .png to .svg everywhere, but the page still renders with .png [19:48] but if I put some garbage at the top of the file, it correctly fails to load [19:58] it worked! [19:58] changing all png and PNG to svg and SVG in my Smokeping.pl works [19:58] it displays a bit big though... [19:59] adding width="$xs{''}" height="$ys{''} in the right place fixes the size... [20:00] example: http://kremvax.acfsys.net/smokeping.cgi?target=Remote.googledns [20:00] mercutio: how does that look on your high-dpi monitor? [20:17] the text looks a bit funny [20:17] i think the font is just crap [20:18] umm rrdtool needs a parameter to generate svg [20:18] oh it is svg [20:18] does it autodetect? [20:19] i should try on my 4k [20:19] erk [20:19] can this font be changed, it's really annoying [20:19] mercutio: you try the proprietary driver? [20:19] it looks fine [20:20] mnathani: nope [20:20] i really don't want to use binary video driver. [20:20]